Title: Nonparametric estimation for optimal dividend barrier with insurance portfolio
Authors: Hiroshi Shiraishi - Keio University (Japan) [presenting]
Zudi Lu - The University of Adelaide (Australia)
Abstract: How an insurance portfolio is used to provide dividend income for insurance company's shareholders is an important problem in application of risk theory, where the premium income as dividends is paid to the shareholders, whenever the surplus attains a level barrier, until the next claim occurs. Under the aggregate claims process taken as a compound Poisson model, we define optimal dividend barrier as the level of the barrier that maximizes the expectation of the discounted dividends until ruin. In the literature, the optimal dividend barrier was derived explicitly under some fixed models concerning claim distribution where parametric estimation is possible. In practice, it may often be hard to provide with the claim distribution parametric model either only from theoretical point of view or from a finite sample, and thus non-parametric estimation is preferred. We consider the non-parametric estimation of the optimal dividend barrier. This study would contribute in practice to decision-making on dividend barrier in the case where a new product is launched or optimality of an existing dividend barrier is tested.